Rapid Development and Evaluation of Shared Control Functions based on an Open Source Physics Simulation for Minimally Invasive Robotic Surgery

Abstract

During the last years the field of medical robotics, especially the number of robotic systems for minimally invasive surgery, has grown. While commercially available systems only provide pure telemanipulation, the aim of current research is to further enhance the surgeon’s performance by adding task-dependent assistive functions based on the concept of shared control. In order to do so, an estimation of the current task state is required. It can be determined from the robot’s current environment. Measuring the required environment information for that is one of the biggest challenges in current research and is not yet fully solved. In this thesis a new idea on solving this problem is introduced. As target system, the minimally invasive robotic surgery system MiroSurge developed at the German Aerospace Center (DLR) was used. In a first step the robot and all of its surrounding components are remodeled in a physics simulation. Afterwards, inside the simulation all kinds of information about the environment can be gathered, including wrench information at the robot’s tip as well as all kinds of position and velocity data. In a next step, the additional information generated from the simulation is used to recognize the currently performed state of a specific task from surgical training. Tools for task representation and state estimation already developed at DLR are used for that. Based on the knowledge about the current task state and the extensive environment information, exemplary assistive functions are implemented for the target task. Their performance is evaluated in a small user study. Finally, the suitability of environment remodeling inside a physics simulation as a tool for rapid development and evaluation of state dependent shared control functions is discussed.